Analyzing the tone of text data

Abstract

Embodiments of the present disclosure relate to analyzing the tone of text data. Aspects of the present invention disclose a method for detecting text input of a user of a social media application and deriving insights into the personality characteristics of the user. The method includes one or more processors that identify text data entries for an interactive Internet-based application. The method also includes determining the tone of the text data entry. The method also includes identifying a publishing account corresponding to the text data entry. The method also includes generating a personality profile corresponding to the publishing account based on the text data entry associated with the publishing account. The method also includes determining the context of the text data entry based on semantic features of the text data entry. The method also includes classifying the personality profile corresponding to the publishing account. The method also includes performing a defined action that prevents participation between the publishing account and the interactive Internet-based application.

Description

Technical Field

[0001] The present invention relates generally to the field of social networking systems, and more particularly to managing comment postings on social media. Background Art

[0002] In recent years, there has been an increasing demand for analyzing large and / or complex data sets using advanced techniques. In particular, Natural Language Processing (NLP) is a subfield of computer science that enables computers to process and analyze large amounts of natural language data. Sentiment analysis utilizes NLP, computational linguistics, and text analysis to extract and analyze subjective information. The basic task of sentiment analysis is to classify the polarity of a given text, where the expressed opinion of a given text is positive, negative, or neutral. Advanced sentiment classification techniques are also able to determine the tone of expression of a given text.

[0003] Neural networks are computing systems modeled after the human brain that provide a framework for many different machine learning algorithms to work together and process complex data inputs. The neural network is first trained, where training involves providing input data and telling the network what the output should be. Neural networks have been used for a variety of tasks (e.g., speech recognition, machine translation, etc.).

[0004] Social media is an interactive, computer-mediated technology that facilitates the creation and sharing of information through virtual communities and networks. User-generated content, such as text posts or comments, photos, videos, and data generated through online interactions, is the lifeblood of social media. Users typically access social media services via web-based technologies on desktop and laptop computers, or download services that provide social media functionality to their mobile devices (e.g., smartphones and tablet computers). Summary of the invention

[0005] Aspects of the present invention disclose a method, computer program product, and system for detecting negative text input by a user in a social media application and delivering an API for deriving personality characteristic insights to an administrator. The method includes identifying, by one or more processors, a text data entry for an interactive Internet-based application. The method also includes determining, by one or more processors, the tone of the text data entry. The method also includes identifying, by one or more processors, a publishing account corresponding to the text data entry. The method also includes generating, by one or more processors, a personality profile corresponding to the identified publishing account based on the text data entry associated with the identified publishing account. The method also includes determining, by one or more processors, the context of the text data entry based on the semantic features of the text data entry. The method also includes classifying, by one or more processors, the personality profile corresponding to the identified publishing account. The method also includes, in response to classifying the personal profile corresponding to the identified publishing account as a first classification, performing, by one or more processors, a defined action that prevents engagement between the identified publishing account and the interactive Internet-based application. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 is a functional block diagram of a data processing environment according to an embodiment of the present invention.

[0007] Figure 2 is a diagram illustrating an embodiment of the present invention Figure 1 A flowchart of the operating steps of a program within a data processing environment for detecting negative text input by a user in a social media application and delivering an API to a management party for deriving personality trait insights.

[0008] Figure 3A is an example depiction of a database object including data corresponding to characteristics of a user according to an embodiment of the present invention.

[0009] Figure 3B is an example depiction of a database object including data corresponding to the tone of text data according to an embodiment of the present invention.

[0010] Figure 3C is an example depiction of a database object including data corresponding to semantic features of text data according to an embodiment of the present invention.

[0011] Figure 3D is an example depiction of a database object including data corresponding to a context of textual data according to an embodiment of the present invention.

[0012] Figure 3Eis an example depiction of a database object including data corresponding to a profile derived from textual data according to an embodiment of the present invention.

[0013] Figure 4 According to an embodiment of the present invention Figure 1 A block diagram of the components of a client device and a server. DETAILED DESCRIPTION

[0014] Embodiments of the present invention allow for the delivery of an application programming interface (API) to a user based on text input to a posting account of a social media application, the API being used to derive personality trait insights associated with a message. Embodiments of the present invention detect and determine the tone of text input to a posting account of a social media application. Embodiments of the present invention determine a profile associated with a posting account based on text input to a posting account. Additional embodiments of the present invention derive the subject matter and context of text input to a posting account to identify topic trends in text input corresponding to one or more posting accounts that interact with content of a domain of a user of a social media application.

[0015] Some embodiments of the present invention recognize that current methods of identifying bad user engagement within the social media domain are primarily methods of categorizing and routing bad text data for review to take certain remedial actions. However, as the practice of sowing discord by posting inflammatory, off-topic, extraneous, or off-topic information in online communities becomes increasingly popular, current methods are inefficient in identifying effective bad user engagement. Various embodiments of the present invention address this problem by utilizing cognitive analysis, tone analysis, and natural language processing (NLP) to categorize posting accounts, verify comments corresponding to users, and derive the context and subject of verified comments to determine profiles and trends of comments within online communities.

[0016] Embodiments of the present invention can improve the efficiency of computer systems by reducing the amount of memory resources utilized to discard irrelevant information. In addition, various embodiments of the present invention can improve the efficiency of network resources by limiting access to posting accounts classified as instigators to reduce the amount of data that the network must transmit.

[0017] Implementation of embodiments of the present invention may take various forms, and exemplary implementation details are discussed below with reference to the drawings.

[0018] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 is a functional block diagram illustrating a distributed data processing environment, generally designated 100 , according to one embodiment of the present invention. Figure 1This merely provides an illustration of one implementation and does not imply any limitation on the environments in which different embodiments may be implemented. Those skilled in the art may make many modifications to the depicted environment without departing from the scope of the invention as described in the claims.

[0019] Various embodiments of the present invention can utilize accessible sources of personal data, which may include personal devices (e.g., client devices 120) social media content and / or publicly available information. For example, embodiments of the present invention may optionally include a privacy component that enables users to opt-in or opt-out of publicizing personal information. The privacy component can support authorized and secure processing of user information such as tracking information and personal information that may have been obtained, maintained and / or accessible. Users may be provided with notifications of the portions of personal information that are collected, as well as the opportunity to opt-in or opt-out of the collection process. Consent can take several forms. Opting-in consent can force users to take affirmative action before data is collected. Alternatively, opting-out consent can force users to take affirmative action before data is collected to prevent the collection of data.

[0020] An embodiment of data processing environment 100 includes client devices 120 and server 130 all interconnected via network 110. In one embodiment, client devices 120 and server 130 communicate via network 110. Network 110 may be, for example, a local area network (LAN), a telecommunications network, a wide area network (WAN) such as the Internet, or any combination thereof, and include wired, wireless, or fiber optic connections. In general, network 110 may be any combination of connections and protocols that will support communication between client devices 120 and server 130, according to embodiments of the present invention. For example, a user of a mobile device (e.g., client device 120) posts a comment on a social media page (e.g., a web application) hosted on a server (e.g., server 130) using the Internet (e.g., network 110).

[0021] Client device 120 may be any device capable of executing computer-readable program instructions. In various embodiments of the present invention, client device 120 may be a workstation, a personal computer, a digital video recorder, a media player, a personal digital assistant, a mobile phone, or any other device capable of executing computer-readable program instructions according to embodiments of the present invention. For example, client device 120 is a mobile device that a user utilizes to respond to a customer review posted on a social media site. According to embodiments of the present invention, client device 120 may include, for example, Figure 4 Components depicted and described in greater detail.

[0022] The client device 120 includes a user interface 122 and an application 124. In various embodiments of the present invention, a user interface is a program that provides an interface between a user of the device and multiple applications resident on the client device. A user interface, such as the user interface 122, refers to information (such as graphics, text, and sound) presented by a program to a user, as well as a control sequence adopted by a user to control the program. There are various types of user interfaces. In one embodiment, the user interface 122 is a graphical user interface. A graphical user interface (GUI) is a user interface that allows a user to interact with an electronic device such as a computer keyboard and mouse through graphical icons and visual indicators such as auxiliary symbols, which is different from a text-based interface, typed command labels, or text navigation. In computing, GUIs were introduced in response to the steep learning curve of command line interfaces that require typing commands on a keyboard. Actions in a GUI are typically performed by directly manipulating graphical elements. In another embodiment, the user interface 122 is a script or application programming interface (API).

[0023] Application 124 is a computer program designed to run on client device 120. Applications are often used to provide users with similar services that are accessed on personal computers (e.g., web browsers, playing music or other media, etc.). In one embodiment, a user utilizes application 124 of client device 120 to access content. For example, application 124 is a web browser of a personal computer, and a user can utilize the web browser to access a social media website. In another embodiment, a user utilizes application 124 of client device 120 to register with review program 200 and define user preferences. For example, application 124 is a web browser of a mobile device, and a user can utilize the web browser to set actions and notification settings for defined actions in response to review program validation of a review.

[0024] In various embodiments of the present invention, server 130 may be a desktop computer, a computer server, or any other computer system known in the art. In some embodiments, server 130 represents a computer system that utilizes clustered computers and components (e.g., a database server computer, an application server computer, etc.) that acts as a single pool of seamless resources when accessed by elements of data processing environment 100. In general, server 130 represents any electronic device or combination of electronic devices that can execute computer-readable program instructions. According to embodiments of the present invention, server 130 may include, for example, Figure 4 Components depicted and described in greater detail.

[0025] In various embodiments, a user authorizes review program 200 to collect and store in storage device 132 information associated with the device and the user who has authorized the collection of information (i.e., the user who has opted in). In one scenario, the owner of client device 120 authorizes review program 200 to collect and store text data (e.g., reviews, customer feedback, text data, etc.) of application 124. In another scenario, the owner of client device 120 authorizes review program 200 to use the collected information of application 124 to perform defined actions on the owner's behalf. In an alternative embodiment, the user opts in to utilize review program 200. For example, the user visits a website associated with review program 200, and the user accepts the terms and conditions of review program 200 as part of the registration process.

[0026] Server 130 includes storage device 132, database 134, and review program 200. Storage device 132 may be implemented using any type of storage device, for example, permanent storage 405, which is capable of storing data that can be accessed and utilized by server 130 and client device 120, such as a database server, a hard drive, or flash memory. In one embodiment, storage device 132 may represent multiple storage devices within server 130. In various embodiments of the invention, storage device 132 stores multiple information, such as one or more instances of database 134. In another embodiment, data processing environment 100 may include additional servers (not shown) that host additional information accessible via network 110.

[0027] The review program 200 can detect negative text input of a publishing account of a social media application based on the text input of the publishing account and deliver an API, which is used to derive personality characteristics insights of the publishing account. In one embodiment, the review program 200 uses the application 124 to identify text data of a web application. For example, the review program 200 parses the comment segmentation of a user on a social media site to detect text input of one or more publishing accounts of the social media site. In another embodiment, the review program 200 determines the tone of the text data of the application 124. For example, the review program 200 uses a neural network to perform language analysis to detect emotions and language tones in the written text of comments on social media websites. In various embodiments of the present invention, the review program 200 can be executed locally on a client device 120 or a server 130.

[0028] In another embodiment, the review program 200 generates a profile score for the publishing account based on the text data associated with the publishing account, and classifies the publishing account based on the publishing account profile score. For example, the review program 200 collects publicly available written text associated with the identifier of the publishing account, and assigns scores to characteristics derived from the publicly available written text. In this example, the review program 200 classifies the publishing account based on the assigned characteristic scores. In addition, the review program 200 can classify the publishing account as an agitator or a moderate with respect to the profile score, tone, and / or characteristics. In another embodiment, the review program 200 determines relevant information from the text data of the application 124, verifies the relevant information, and generates an object based on the relevant information. For example, the relevant information may include characteristics, scores, tone, etc. derived from the text input of the publishing account.

[0029] Figure 2 1 is a flowchart depicting the operating steps of a review program 200 according to an embodiment of the present invention, the review program 200 being a program for detecting negative text input to a publishing account of a social media application and delivering an API to a user for deriving personality trait insights. In one embodiment, the review program 200 is launched to analyze an existing set of text data. For example, the review program 200 is automatically launched in response to a publishing account leaving a comment under a user's social media post. In another embodiment, the review program 200 continuously monitors the application 124 for text data. For example, the review program 200 monitors notifications of a social media application to detect a comment notification, and the review program 200 is launched in response to detecting a comment notification.

[0030] In step 202, the review program 200 identifies text data of a web application. In one embodiment, the review program 200 extracts data of the application 124 to identify text data entries of the application 124 (e.g., an interactive Internet-based application). For example, the review program 200 retrieves structured and unstructured text data of a user-specific profile of a defined social media medium. In this example, the review program 200 uses text analysis (e.g., text data mining) to detect comments in the user-specific profile of the defined social media medium. In another embodiment, the review program 200 monitors the application 124 to detect a notification indicating that a text data entry has been made using an interactive Internet-based application. For example, the review program 200 monitors a notification application (e.g., application 124) of a mobile device (e.g., client device 120) and detects a notification corresponding to a profile of a user associated with the social media application. In this example, the review program 200 uses natural language processing (NLP) to determine whether the notification corresponds to a comment (e.g., a text data entry).

[0031] In step 204, the review program 200 determines the tone of the text data. In one embodiment, the review program 200 uses a cognitive model to determine the tone of the text data entry of the application 124. For example, the review program 200 can use supervised learning (e.g., a support vector machine (SVM)) to train a machine learning model (e.g., a neural network) to identify the tone of the comment. In addition, the review program 200 generates a classifier for each category, where the training set consists of a set of test comments (positive labels) and their complements (negative labels) in the category, and a given test comment maps each classifier individually. In this example, the review program 200 uses features of several categories (e.g., N-tuples (unit groups, bigrams, and trigrams), punctuation, emoticons, curse words, greetings (such as "hello", "hi" and "thank you"), and sentiment polarity, etc.) to classify the tone of comments of user-specific profiles of a defined social media medium. In another example, the review program 200 uses a one-over-all (OVR) paradigm to independently train models for one or more tones, where the paradigm uses reviews of each category as positive samples and all other reviews as negative samples, and identifies the tone predicted with at least 0.5 probability as the final tone.

[0032] In another example, the review program 200 inputs a string corresponding to a review into a machine learning model that provides a score (e.g., in the range of 0.5 to 1, where 1 is a larger value) for one or more emotion and / or language tone classifications (e.g., anger, fear, happiness, sadness, analytical, confident, tentative, etc.) and assigns an identifier (ID) to the review. In this example, the review program 200 compares the score of the one or more emotion and / or language tone classifications with a defined threshold (e.g., 0.75, which indicates a high likelihood of perceiving the tone) associated with a target F1 score of the machine learning model. In addition, the review program 200 assigns an ID to the tone having a score equal to or exceeding the defined threshold and / or the tone having the highest score. In addition, the review program 200 stores the review and data corresponding to the review (e.g., ID, score, etc.) in a database of a server (e.g., server 130).

[0033] In decision step 206, the review program 200 determines whether the tone of the text data is negative. In one embodiment, the review program 200 determines the tone based on the text data entry of the application 124. For example, the review program 200 utilizes NLP and cognitive linguistic techniques to classify the polarity and tone of the comments of the user-specific profile of the defined social media medium based on the text data of the comments. In this example, the review program 200 parses the comments of the user-specific profile to detect words and tones that are generally associated with polarity (e.g., negative, neutral, positive, etc.). In addition, based on how the detected words are related to concepts in a specific domain (e.g., customer service, product support, problem reporting, etc.), the detected words associated with polarity are ranked and assigned integers in the range of -5 to +5 (i.e., the most negative to the most positive), which allows the rating of the word to be adapted according to the concept in the context of the specified environment. In addition, the review program 200 assigns a score to the identified polarity and / or expression tone of the comment based on the ranking of the detected words included in the comment.

[0034] If the review program 200 determines that the tone of the text data entry is positive (decision step 206, "no" branch), the review program 200 proceeds to identify the text data entry for the application 124 (step 202). In one scenario, if the review program 200 parses the review and detects a word that is typically associated with positive polarity and tone (e.g., assigned the integer 3), the review program 200 proceeds to use text analysis (e.g., text data mining) to detect reviews in a user-specific profile of a defined social media medium.

[0035] At step 208, review program 200 identifies the posting account that generated the text data. More specifically, in response to review program 200 determining that the tone of the text data entry is negative (decision step 206, "yes" branch), review program 200 identifies the posting account as the source of the review. In one scenario, if review program 200 parses the review and detects words that are typically associated with negative polarity and tone (e.g., assigned an integer of -3), review program 200 extracts an identifier corresponding to the posting account that is the source of the review.

[0036] In one embodiment, the review program 200 uses NLP to determine the identifier of the posting account corresponding to the text data entry of the defined social media media. For example, the review program 200 retrieves the structured and unstructured text data of the posting account from the publicly available social media media (e.g., social media public profile). In this example, the review program 200 uses text analysis (e.g., text data mining) to identify the user ID (e.g., user name, uniform resource locator URL, etc.) of the posting account in the text data of the comment of the user-specific profile. In addition, the review program 200 stores the user ID and the comment in the database of the server.

[0037] At step 210, the review program 200 generates a profile of the posting account. In one embodiment, the review program 200 generates a set of data corresponding to the identifier of the posting account using the text data of the application 124. For example, the review program 200 uses a machine learning algorithm to create a characteristic profile that corresponds to the extracted user ID of the posting account associated with the text data of the publicly available profile of the posting account. Typically, the review program 200 can generate a profile using multiple text data entries corresponding to the posting account. In this example, the review program 200 uses an open-vocabulary approach to train the machine learning algorithm using the scores from a survey conducted between multiple posting accounts and the derived profile data of one or more posting accounts of a defined social media medium.

[0038] In addition, the machine learning model includes five (5) personality traits that represent user engagement (e.g., friendliness, conscientiousness, extroversion, emotional range, openness, etc.), twelve (12) needs that represent aspects of the product that resonate with the author of the review (e.g., excitement, harmony, curiosity, ideals, intimacy, self-expression, freedom, love, practicality, stability, challenge, structure, etc.), and five (5) values ​​that represent motivating factors that influence user decisions (e.g., self-transcendence, tradition, hedonism, self-improvement, excitement, etc.). In addition, the review program 200 uses the machine learning model to generate scores corresponding to the identified personality traits and values, where scores above the mean of 0.5 on a scale of zero (0) to one (1) indicate a trend for the trait that is above average, and scores at or above 0.75 indicate easily discernible aspects of the trait.

[0039] In another example, the review program 200 retrieves publicly available structured and unstructured text data corresponding to the extracted user ID, and tokenizes the comments of the text data to form a representation in an n-dimensional space. In addition, the review program 200 uses an unsupervised learning algorithm to obtain a vector representation for words in the input text (e.g., the words of the comment). In this example, the review program 200 feeds the input text into a machine learning algorithm, which generates a normalized score for the input text (e.g., the comment) by comparing the original score with the results from the sample population, and the review program 200 uses the score to infer the personality profile of the publishing account including personality, needs, and values ​​characteristics. The review program 200 reports the percentage of personality, needs, and values ​​characteristics as a double-precision floating point number in the range of zero (0) to one (1) based on the quality inferred from the input text. In addition, the percentile 0.64980796071382 for the personality characteristic indicates that the score of the publishing account for the characteristic is at the 65th percentile.

[0040] Figure 3A Depicted is a profile object 300, which is an example of a database object created by the review program 200 that includes various fields. The profile object 300 includes the fields: trait_id, big five, category, percentile, and score. "Trait_id" is a string that is the unique ID of the characteristic to which the result belongs (e.g., the big five personality dimensions). "Big five" is a personality characteristic that represents the user's participation. "Category" is a string indicating the category of the characteristic (e.g., personality, needs, values, etc.), where personality is a recursive array of trait objects describing the big five dimensions and aspects inferred from the input text, needs is an array of trait objects describing the needs inferred from the input text, and values ​​is an array of trait objects describing the values ​​inferred from the input text. "Percentile" is a normalized percentile score for the characteristic. "Score" is a raw score for the characteristic. In an example embodiment, the review program 200 utilizes the input text of the application 124 (e.g., a social media comment) to generate a profile object 300 corresponding to the user ID of the publishing account that submitted the input text.

[0041] At step 212, the review program 200 categorizes the activity of the profile of the posting account. In one embodiment, the review program 200 categorizes the profile of the posting account based on a generated data set corresponding to the profile of the posting account. For example, the review program 200 compares the personality, needs, and values ​​characteristics of the generated personality profile of the posting account with the personality profile of an agitator or an opponent to categorize the posting account. In this example, the review program 200 compares the percentiles of the characteristics of the profiles of the posting account and the agitator / opponent to determine whether the characteristics of the profiles match, wherein a percentile above the mean of 0.5 of the range of zero (0) to one (1) for a particular characteristic of the two profiles indicates a match.

[0042] Figure 3B Depicted is a tone object 400, which is an example of a database object created by the review program 200 that includes various fields. The tone object 400 includes the fields: tone_id, type, classification, comment, and handle. "Tone_id" is a unique, non-localized identifier of the identified tone of the input text. "Type" is the polarity identified in the input text. "Classification" is a personality identifier of the publishing account associated with the source of the input text. "Comment" is a string representing the input text. "Handle" is the social media user ID of the publishing account associated with the source of the input text.

[0043] In an example embodiment, the review program 200 populates the mood object 400 with the generated and extracted data. In this example, the review program 200 extracts handles (e.g., user IDs) and comments from the input text of the application 124. In addition, the review program 200 utilizes a machine learning algorithm to determine the mood of the input text. In addition, the review program 200 incrementally assigns a mood_id to the received input text.

[0044] In step 214, the comment program 200 determines relevant information from the text data. In various embodiments of the present invention, the comment program 200 identifies and analyzes semantic features of the text data (e.g., categories, concepts, keywords, etc.) to determine relevant information (e.g., context, topic, etc.) of the comments of the publishing account of the social media medium. In one embodiment, the comment program 200 uses natural language understanding (NLU) to determine relevant information of the text data entry of the application 124. For example, the comment program 200 uses a semantic parser to convert the comments of the publishing account of the social media media into a primary logical structure and identifies the expected semantics of the primary logical structure. In this example, the comment program 200 determines the context and topic of the comment based on the semantic features of the comment. In addition, the comment program 200 stores the determined context and topic of the comment in a database of the server.

[0045] Figure 3CA context object 500 is depicted, which is an example of a database object created by the review program 200 that includes various fields. Object 500 includes the fields: context_id, category, and concept. "Context_id" is a string that is a unique ID of the identified semantic feature of the text input. "Category" is the taxonomy of the input text, which includes a five-level classification hierarchy. "Concept" is a high-level idea of ​​the input text. For example, a research paper on deep learning may return the concept "artificial intelligence", although the term is not mentioned. In an example embodiment, the review program 200 utilizes NLU to extract semantic features (e.g., context, concept, etc.) from the input text of the application 124 (e.g., social media comments) to populate the context object 500.

[0046] At step 216, the review program 200 validates the text data. In one embodiment, the review program 200 utilizes NLU and the determined relevant information to determine whether the text data entry of the application 124 is valid user feedback. For example, the review program 200 retrieves the context and subject of the comment of the posting account of the social media medium (determined in step 214). In addition, by matching the context and subject of the comment with the context and subject of one or more related comments of the posting account using NLU technology, the review program 200 identifies one or more related comments of the posting account. In this example, the review program 200 compares the percentile of the characteristic of the one or more related comments with the percentile of the characteristic of the agitator / opponent profile to determine whether the characteristic of the one or more related comments matches, wherein a percentile above the mean of 0.5 in the range of zero (0) to one (1) for a specific characteristic of the one or more related comments indicates a match.

[0047] In one scenario, if the review program 200 determines that one or more relevant comments match the agitator / detractor profile, the review program 200 stores the comments and related data (e.g., date, characteristics, context, etc.) in a database object for user feedback (e.g., an agitator / detractor table). In another scenario, if the review program 200 determines that one or more relevant comments do not match the agitator / detractor profile, the review program 200 stores the comments and related data in a database object for user feedback (e.g., a feedback table).

[0048] Figure 3DDepicted are feedback objects 600 and agitator objects 700, which are examples of database objects created by review program 200 that include various fields. Feedback objects 600 and agitator objects 700 include fields: tone_id, trait_id, context_id, and date. In an example embodiment, review program 200 extracts the tone_id, trait_id, and context_id of a review (e.g., input text) of application 124 from profile object 300, tone object 400, and context object 500, respectively, to populate feedback objects 600 and agitator objects 700. In addition, review program 200 extracts the input date from metadata of application 124 to populate the date field of feedback object 600 or agitator object 700.

[0049] At step 218, the review program 200 performs the defined action. In one embodiment, the review program 200 detects a set of conditions and performs the defined action using the application 124. For example, the review program 200 detects a post of a user of a social media medium, receives an agitator comment, and transmits a message to an administrator of a user-specific profile corresponding to the post. In this example, the message may include a custom database object created by the review program 200 that includes data associated with the agitator comment. In another example, the review program 200 detects a profile of a user of a social media medium, receives an agitator comment, and based on preferences provided by an administrator and / or user of the profile, the review program 200 may delete the agitator comment and / or delete the agitator's user ID from the social media medium. In yet another example, the review program 200 may prevent the agitator's user ID from interacting (e.g., commenting) with the user's profile.

[0050] At decision step 220, review program 200 determines whether a trend exists in the verified text data. In one embodiment, review program 200 determines whether the relevant information of the verified text data entry exceeds a user-defined number of matches of the relevant information of the data set of verified text data entries in database 134. For example, review program 200 monitors a database (e.g., database 134) to detect storage of verified reviews (e.g., verified text data entries) in a database object of a server (e.g., server 140). In this example, review program 200 determines whether the context (e.g., relevant information) of the verified review matches the context of the review of the database object of the server. In addition, review program 200 determines whether the number of matches exceeds a defined threshold number of matches to establish a trend in the reviews of the database object.

[0051] If the review program 200 determines that the information associated with the verified text data entry does not exceed the user-defined number of matches of the information associated with the data set of verified text data entries of database 134 (decision step 220, "no" branch), the review program 200 proceeds to determine whether the text data entry of application 124 is valid user feedback (step 216). For example, if the review program 200 determines that the number of context matches does not exceed a defined threshold number of matches to establish trends in reviews of database objects, the review program 200 proceeds to determine whether the review matches an agitator / detractor profile.

[0052] At step 222, review program 200 provides a profile alert. More specifically, in response to review program 200 determining that the information associated with the verified text data entry exceeds a user-defined number of matches of the information associated with the data set of verified text data entries of database 134 (decision step 220, "yes" branch), review program 200 returns the generated database object of database 134. For example, if review program 200 determines that the number of context matches exceeds a defined threshold number of matches to establish trends in the reviews of the database object, review program 200 provides the custom database object created by review program 200 (at step 218), the custom database object including data associated with the agitator review.

[0053] Figure 3E An alert object 800 is depicted, which is an example of a database object created by the review program 200 that includes various fields. The alert object 800 includes the fields: tone_id, type, comment, handle, context, concept, agitator warning, topic warning, and context warning. In an example embodiment, the review program 200 extracts the tone_id, type, comment, handle, context, and concept of the review (e.g., input text) of the application 124 from the profile object 300, the tone object 400, and the context object 500, respectively, to populate the alert object 800. In this example, the review program 200 utilizes the above-described techniques (in steps 210 and 212) to determine an agitator warning for the input text. In addition, the review program 200 utilizes the NLU and data from the context object 500 to determine whether the topic and context of the input text are in-domain (i.e., the derived topic and context of the review are not in the specific domain associated with the administrator's profile). Additionally, review program 200 returns alert object 800 to the administrator in response to the query and / or review program 200 identifies trends in the data set of validated text data entries of database 134 .

[0054] Figure 4 1 depicts a block diagram of components of a client device 120 and a server 130 according to an illustrative embodiment of the present invention. It should be understood that Figure 4This merely provides an illustration of one embodiment and does not imply any limitations to the environments in which different embodiments may be implemented.Many modifications to the depicted environments may be made.

[0055] Figure 4 The system includes (multiple) processors 401, cache 403, memory 402, permanent storage 405, communication unit 407, (multiple) input / output (I / O) interface 406, and communication structure 404. The communication structure 404 provides communication between the cache 403, the memory 402, the permanent storage 405, the communication unit 407, and (multiple) input / output (I / O) interface 406. The communication structure 404 can be implemented using any architecture designed to transfer data and / or control information between processors (such as microprocessors, communication and network processors, etc.), system memory, peripheral devices, and any other hardware components within the system. For example, the communication structure 404 can be implemented using one or more buses or crossbar switches.

[0056] Memory 402 and permanent storage 405 are computer-readable storage media. In this embodiment, memory 402 includes random access memory (RAM). In general, memory 402 may include any suitable volatile or non-volatile computer-readable storage media. Cache 403 is a fast memory that enhances the performance of processor(s) 401 by saving recently accessed data from memory 402 and data near recently accessed data.

[0057] Program instructions and data (e.g., software and data 410) for practicing embodiments of the present invention may be stored in permanent storage 405 and memory 402 for execution by one or more of the corresponding processors 401 via cache 403. In one embodiment, permanent storage 405 includes a magnetic hard drive. Alternatively, or in addition to a magnetic hard drive, permanent storage 405 may include a solid-state hard drive, a semiconductor memory device, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), flash memory, or any other computer-readable storage medium capable of storing program instructions or digital information.

[0058] The media used by the permanent storage 405 may also be removable. For example, a removable hard drive may be used for the permanent storage 405. Other examples include optical and magnetic disks, thumb drives, and smart cards that are inserted into a drive for transfer to another computer-readable storage medium that is also part of the permanent storage 405. Software and data 410 may be stored in the permanent storage 405 for access and / or execution by one or more of the corresponding processor(s) 401 via the cache 403. With respect to the client device 120, the software and data 410 includes data for the application 124. With respect to the server 130, the software and data 410 includes data for the review program 200 and the storage device 132.

[0059] In these examples, communication unit 407 provides communications with other data processing systems or devices. In these examples, communication unit 407 includes one or more network interface cards. Communication unit 407 can provide communications by using one or both of physical and wireless communication links. Program instructions and data (e.g., software and data 410) for practicing embodiments of the present invention can be downloaded to permanent storage device 405 through communication unit 407.

[0060] (Multiple) I / O interface 406 allows input and output of data with other devices that may be connected to each computer system. For example, (multiple) I / O interface 406 can provide connection to (multiple) external device 408, such as a keyboard, a keypad, a touch screen, and / or some other suitable input device. (Multiple) external device 408 can also include portable computer readable storage media, such as, for example, a thumb drive, a portable optical or magnetic disk, and a memory card. Program instructions and data (e.g., software and data 410) for practicing embodiments of the present invention can be stored on such portable computer readable storage media and can be loaded onto permanent storage device 405 via (multiple) I / O interface 406. (Multiple) I / O interface 406 is also connected to display 409.

[0061] Display 409 provides a mechanism for displaying data to a user and may be, for example, a computer monitor.

[0062] The programs described herein are identified based on the applications for which they are implemented in specific embodiments of the invention. However, it should be understood that any specific program terminology used herein is merely for convenience, and therefore, the present invention should not be limited to use in any specific application identified and / or implied by such terminology.

[0063] The description of various embodiments of the present invention has been presented for the purpose of illustration and is not intended to be exhaustive or to limit the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the present invention. The terms used herein are selected to best explain the principles of the embodiments, practical applications, or technical improvements to technologies found on the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method, include: identifying, by one or more processors, textual data entries for an interactive internet-based application; determining, by one or more processors, that the tone of the text data entry includes a negative tone; identifying, by one or more processors, a publishing account corresponding to the text data entry; generating, by one or more processors, a personality profile corresponding to the identified posting account based on the text data entries associated with the identified posting account; determining, by one or more processors, a context of the text data entry based on semantic features of the text data entry; categorizing, by one or more processors, the personality profile corresponding to the identified publishing account into a first category; validating, by one or more processors, the text data entry based on the first classification and the context, wherein validating comprises determining that the negative tone of the text data entry is a valid use of feedback and does not include a trend; and In response to classifying the personality profile corresponding to the identified publishing account as a first classification and verifying the text data entry, a defined action is delayed that prevents engagement between the identified publishing account and the interactive Internet-based application.

2. The method according to claim 1, further comprising: include: identifying a second text data entry for an interactive Internet-based application, the second text data entry comprising a second context; determining that the tone of the second text data item includes a negative tone; verifying, by one or more processors, the second text data entry of the identified posting account; as well as determining, by one or more processors, that the second context of the second verified text data entry exceeds a defined threshold for a data set including two or more verified text data entries; as well as In response to determining that the second context of the second verified text data entry exceeds the defined threshold for the data set, the defined action is performed.

3. The method according to claim 2, further comprising: include: generating, by one or more processors, an object for a database based at least in part on the text data entry, the context of the text data entry, and the personality profile corresponding to the identified publishing account; as well as In response to determining that the context of the validated text data entry exceeds a defined threshold for the data set containing two or more validated text data entries, the object of the database is returned by one or more processors.

4. The method of claim 1 , wherein generating the personality profile corresponding to the identified posting account based on the text data entry associated with the identified posting account further comprises: include: retrieving, by one or more processors, structured and unstructured text data of the interactive Internet-based application associated with the identified posting account, wherein the structured and unstructured text data includes one or more comments of the identified posting account; tokenizing, by one or more processors, the one or more comments of the posting account identified; inputting, by one or more processors, the tokenized one or more reviews into a machine learning algorithm; as well as A set of data for the identified publishing account is generated by one or more processors based at least in part on a normalized score of the machine learning algorithm for the tokenized one or more comments, wherein the set of data includes the normalized scores of one or more personality traits, value traits, and need traits.

5. The method according to claim 1, wherein the context of the text data entry is determined based on the semantic features of the text data entry. include: parsing, by one or more processors, the textual data entry for the identified posting account; as well as One or more semantic features of the text data entry of the identified publishing account are identified by one or more processors, wherein the one or more semantic features include at least a category, a concept, and a keyword of the text data entry.

6. The method of claim 1, wherein the personality profile corresponding to the identified publishing account is further classified. include: extracting, by one or more processors, one or more characteristics of the personality profile of the identified publishing account; comparing, by one or more processors, the extracted one or more characteristics of the personality profile to one or more characteristics of a set of data corresponding to a personality profile of a user tagged as a first classification; determining, by one or more processors, that the personality profile of the publishing account includes more than a defined threshold number of characteristics of the personality profile of the user tagged as the first classification, wherein the determination is based on the comparison; as well as In response to determining that the personality profile of the publishing account includes characteristics of the personality profile of the user marked as the first classification that are higher than the defined threshold number, wherein the determination is based on the comparison, marking the personality profile of the publishing account as corresponding to the first classification by one or more processors.

7. The method of claim 1, wherein the defined action is selected from the group consisting of: transmitting a message, deleting a user ID, blocking a user ID, or removing the text data entry.

8. A computer program product, include: One or more computer-readable storage media and program instructions stored on the one or more computer-readable storage media, the program instructions comprising program instructions for performing the actions of the method according to any one of claims 1 to 7.

9. A computer system, include: one or more computer processors; one or more computer-readable storage media; as well as Program instructions are stored on the computer-readable storage medium for execution by at least one of the one or more processors, the program instructions being used to perform actions of the method according to any one of claims 1 to 7.

10. A computer system comprising modules for executing the steps of the method according to any one of claims 1 to 7.

Images